One such pioneering artificial intelligence tool, Delphi-2M, has recently burst into the scene. Today, it can predict an individual’s risk for more than 1,000 diseases. Originally created by a team of University of Pittsburgh and Yale University researchers, this innovative tool leverages detailed, anonymized patient data from two very different healthcare systems. The AI model was fitted with extreme care. It used data from 400,000 people in the UK Biobank study and 1.9 million patients in the Danish national patient registry.
Delphi-2M is better suited for modeling human disease progression at a large scale. In addition, it forecasts far into the future – as far out as 10 years in advance. The tool rewires anonymized patient records to track their health changes over time. Instead of foretelling doom, it introduces hazards as probabilities, similar to meteorological reports. For example, it could say there is a 70% probability of getting a certain health condition over a given period.
This AI-enabled tool provides personalized risk scores for individuals based on their previous disease history. It then produces predictions of unmatched precision and accuracy that beat current single-disease state-of-the-art models. This advancement signifies a major step toward personalized healthcare, allowing both patients and clinicians to understand potential future health challenges.
Prof. Moritz Gerstung emphasized the significance of this innovation, stating, “This is the beginning of a new way to understand human health and disease progression.” He added that generative models such as Delphi-2M would one day be able to customize healthcare requirements on an enormous, societal scale.
Beyond risk identification, the AI-powered tool offers practical, evidence-based guidance for addressing and minimizing each identified risk. Tomas Fitzgerald highlighted the predictive nature of the model, saying, “Our AI model learns those patterns and can forecast future health outcomes.” Such insights empower clinicians to present patients with clear options: “Here’s four major risks that are in your future and here’s two things you could do to really change that,” as described by an unnamed source.
Given its promise to greatly increase patient outcomes, experts expect to see Delphi-2M in widespread use in clinical environments in the near future. UK bioinformatician Ewan Birney was cautiously optimistic about its rollout into medical practice in the coming few years. He said that this tool would allow healthcare providers to better predict patient needs.
The generative capabilities of Delphi-2M further enable sampling of synthetic future health trajectories. Fitzgerald added, “Delphi-2M’s generative nature enables sampling of synthetic future health trajectories, providing meaningful estimates of potential disease burden for up to 20 years.”